| 1 | use crate::block::{BlockCell, BlockRange, ClientID, Item, ItemPtr, GC, ID}; |
| 2 | use crate::slice::ItemSlice; |
| 3 | use crate::types::TypePtr; |
| 4 | use crate::utils::client_hasher::ClientHasher; |
| 5 | use crate::*; |
| 6 | use std::collections::hash_map::Entry; |
| 7 | use std::collections::HashMap; |
| 8 | use std::hash::BuildHasherDefault; |
| 9 | use std::ops::{Index, IndexMut, Range, RangeInclusive}; |
| 10 | use std::vec::Vec; |
| 11 | |
| 12 | /// A resizable list of blocks inserted by a single client. |
| 13 | #[derive(PartialEq, Default)] |
| 14 | pub(crate) struct ClientBlockList { |
| 15 | list: Vec<BlockCell>, |
| 16 | } |
| 17 | |
| 18 | struct SquashBlockRange { |
| 19 | range: Range<usize>, |
| 20 | gc_block: bool, |
| 21 | } |
| 22 | |
| 23 | impl ClientBlockList { |
| 24 | pub fn clock(&self) -> u32 { |
| 25 | let len = self.list.len(); |
| 26 | if len == 0 { |
| 27 | 0 |
| 28 | } else { |
| 29 | match &self.list[len - 1] { |
| 30 | BlockCell::GC(gc) => gc.end + 1, |
| 31 | BlockCell::Block(block) => block.id.clock + block.len, |
| 32 | } |
| 33 | } |
| 34 | } |
| 35 | |
| 36 | pub(crate) fn get(&self, index: usize) -> Option<&BlockCell> { |
| 37 | self.list.get(index) |
| 38 | } |
| 39 | |
| 40 | /// Given a block's identifier clock value, return an offset under which this block could be |
| 41 | /// found using binary search algorithm, or a index under which this block should be inserted. |
| 42 | pub(crate) fn find_pivot(&self, clock: u32) -> Option<usize> { |
| 43 | let mut left = 0; |
| 44 | let mut right = self.list.len() - 1; |
| 45 | let mut block = &self[right]; |
| 46 | let (mut start, mut end) = block.clock_range(); |
| 47 | if start == clock { |
| 48 | // a common case is to just append a block at the end, so check first if we can do that |
| 49 | Some(right) |
| 50 | } else { |
| 51 | let mut mid = ((clock / end) * right as u32) as usize; |
| 52 | while left <= right { |
| 53 | block = &self[mid]; |
| 54 | (start, end) = block.clock_range(); |
| 55 | if start <= clock { |
| 56 | if clock <= end { |
| 57 | return Some(mid); |
| 58 | } |
| 59 | left = mid + 1; |
| 60 | } else { |
| 61 | right = mid - 1; |
| 62 | } |
| 63 | mid = (left + right) / 2; |
| 64 | } |
| 65 | |
| 66 | None |
| 67 | } |
| 68 | } |
| 69 | |
| 70 | /// Attempts to find a Block which contains given clock sequence number within current block |
| 71 | /// list. Clocks are considered to work in left-side inclusive way, meaning that block with |
| 72 | /// an ID (<client-id>, 0) and length 2, with contain all elements with clock values |
| 73 | /// corresponding to {0,1} but not 2. |
| 74 | fn get_block(&self, clock: u32) -> Option<&BlockCell> { |
| 75 | let idx = self.find_pivot(clock)?; |
| 76 | Some(&self[idx]) |
| 77 | } |
| 78 | |
| 79 | fn get_block_mut(&mut self, clock: u32) -> Option<&mut BlockCell> { |
| 80 | let idx = self.find_pivot(clock)?; |
| 81 | Some(&mut self[idx]) |
| 82 | } |
| 83 | |
| 84 | /// Pushes a new block at the end of this block list. |
| 85 | fn push(&mut self, cell: BlockCell) { |
| 86 | self.list.push(cell); |
| 87 | } |
| 88 | |
| 89 | /// Inserts a new block at a given `index` position within this block list. This method may |
| 90 | /// panic if `index` is greater than a length of the list. |
| 91 | pub(crate) fn insert(&mut self, index: usize, cell: BlockCell) { |
| 92 | self.list.insert(index, cell); |
| 93 | } |
| 94 | |
| 95 | /// Returns a number of blocks stored within this list. |
| 96 | pub fn len(&self) -> usize { |
| 97 | self.list.len() |
| 98 | } |
| 99 | |
| 100 | pub fn iter(&self) -> ClientBlockListIter<'_> { |
| 101 | ClientBlockListIter(self.list.iter()) |
| 102 | } |
| 103 | |
| 104 | pub fn iter_mut(&mut self) -> ClientBlockListIterMut<'_> { |
| 105 | ClientBlockListIterMut(self.list.iter_mut()) |
| 106 | } |
| 107 | |
| 108 | /// Attempts to squash multiple blocks within the given range of indices. |
| 109 | /// For each block in the `indices_range`, it will check if the block can be squashed with its left neighbor. |
| 110 | /// If consecutive blocks are squashable, they are tracked in a range and processed in bulk to compact |
| 111 | /// the list efficiently. The function supports both GC and Block cells. |
| 112 | /// |
| 113 | /// - For GC blocks: If blocks are consecutive, the range is extended and squashing is deferred until |
| 114 | /// all squashable blocks are identified. |
| 115 | /// |
| 116 | /// - For Block cells: The function attempts to squash the contents of the right block into the left block. |
| 117 | /// If successful, it tracks the blocks to be removed and rewires references in the parent node if necessary. |
| 118 | /// Block cells currently don't support range compaction due to the complexity of squashing Blocks. |
| 119 | /// |
| 120 | /// The function processes all blocks in reverse order (from the end of the range to the start), |
| 121 | /// compacts the list by removing squashed blocks, and updates references for any parent-child relationships |
| 122 | /// affected by the squashing. |
| 123 | /// |
| 124 | /// # Arguments |
| 125 | /// * `indices_range` - A range of indices, where each index represents a block in the list to be examined |
| 126 | /// for squashing. The range must be non-empty (`start` must be <= `end`). |
| 127 | /// |
| 128 | /// # Panics |
| 129 | /// * Panics if `indices_range.start()` is greater than `indices_range.end()`. |
| 130 | /// |
| 131 | pub(crate) fn squash_left_range_compaction(&mut self, indices_range: RangeInclusive<usize>) { |
| 132 | assert!(indices_range.start() <= indices_range.end()); |
| 133 | let mut squash_intervals: Vec<SquashBlockRange> = Vec::new(); |
| 134 | |
| 135 | for right_index in indices_range.rev() { |
| 136 | let (l, r) = self.list.split_at_mut(right_index); |
| 137 | let left = &mut l.last_mut().unwrap(); |
| 138 | let right = &mut r[0]; |
| 139 | |
| 140 | match (left, right) { |
| 141 | (BlockCell::GC(_), BlockCell::GC(_)) => { |
| 142 | let mut extended = false; |
| 143 | match squash_intervals.last_mut() { |
| 144 | Some(last_range) if last_range.gc_block => { |
| 145 | // Extend if consecutive |
| 146 | if last_range.range.start - 1 == right_index { |
| 147 | last_range.range.start = right_index; |
| 148 | extended = true; |
| 149 | } |
| 150 | } |
| 151 | _ => {} |
| 152 | } |
| 153 | |
| 154 | if !extended { |
| 155 | // Add new range if no consecutive block found |
| 156 | squash_intervals.push(SquashBlockRange { |
| 157 | range: Range { |
| 158 | start: right_index, |
| 159 | end: right_index, |
| 160 | }, |
| 161 | gc_block: true, |
| 162 | }); |
| 163 | } |
| 164 | } |
| 165 | (BlockCell::Block(left), BlockCell::Block(right)) => { |
| 166 | let mut left = ItemPtr::from(left); |
| 167 | let right = ItemPtr::from(right); |
| 168 | if left.try_squash(right) { |
| 169 | // Merge right into left Blocks one by one. |
| 170 | squash_intervals.push(SquashBlockRange { |
| 171 | range: Range { |
| 172 | start: right_index, |
| 173 | end: right_index, |
| 174 | }, |
| 175 | gc_block: false, |
| 176 | }); |
| 177 | } |
| 178 | } |
| 179 | _ => { /* cannot squash incompatible types */ } |
| 180 | } |
| 181 | } |
| 182 | |
| 183 | for squash_range in &squash_intervals { |
| 184 | let start_idx = squash_range.range.start; |
| 185 | let end_idx = squash_range.range.end; |
| 186 | assert!(start_idx <= end_idx); |
| 187 | |
| 188 | let (left_slice, right_slice) = self.list.split_at_mut(end_idx); |
| 189 | |
| 190 | // The start_idx - 1 element is the one want to squash into. |
| 191 | let left = &mut left_slice[start_idx - 1]; |
| 192 | let right = &right_slice[0]; |
| 193 | |
| 194 | match (left, right) { |
| 195 | (BlockCell::GC(left), BlockCell::GC(right)) => { |
| 196 | left.end = right.end; |
| 197 | } |
| 198 | (BlockCell::Block(left), BlockCell::Block(right)) => { |
| 199 | let left = ItemPtr::from(left); |
| 200 | let right = ItemPtr::from(right); |
| 201 | if let Some(key) = right.parent_sub.as_deref() { |
| 202 | if let TypePtr::Branch(mut parent) = right.parent { |
| 203 | if let Some(e) = parent.map.get_mut(key) { |
| 204 | if right == *e { |
| 205 | *e = ItemPtr::from(left); |
| 206 | } |
| 207 | } |
| 208 | } |
| 209 | } |
| 210 | } |
| 211 | _ => { /* cannot squash incompatible types */ } |
| 212 | } |
| 213 | |
| 214 | // Finally, remove the BlockCells in bulk. |
| 215 | self.list.drain(start_idx..=end_idx); |
| 216 | } |
| 217 | } |
| 218 | |
| 219 | /// Attempts to squash block at a given `index` with a corresponding block on its left side. |
| 220 | /// If this succeeds, block under a given `index` will be removed, and its contents will be |
| 221 | /// squashed into its left neighbor. In such case a squash result will be returned in order to |
| 222 | /// later on rewire left/right neighbor changes that may have occurred as a result of squashing |
| 223 | /// and block removal. |
| 224 | pub(crate) fn squash_left(&mut self, index: usize) { |
| 225 | let (l, r) = self.list.split_at_mut(index); |
| 226 | let left = &mut l[index - 1]; |
| 227 | let right = &mut r[0]; |
| 228 | match (left, right) { |
| 229 | (BlockCell::GC(left), BlockCell::GC(right)) => { |
| 230 | left.end = right.end; |
| 231 | self.list.remove(index); |
| 232 | } |
| 233 | (BlockCell::Block(left), BlockCell::Block(right)) => { |
| 234 | let mut left = ItemPtr::from(left); |
| 235 | let right = ItemPtr::from(right); |
| 236 | if left.try_squash(right) { |
| 237 | if let Some(key) = right.parent_sub.as_deref() { |
| 238 | if let TypePtr::Branch(mut parent) = right.parent { |
| 239 | if let Some(e) = parent.map.get_mut(key) { |
| 240 | if right == *e { |
| 241 | *e = ItemPtr::from(left); |
| 242 | } |
| 243 | } |
| 244 | } |
| 245 | } |
| 246 | self.list.remove(index); |
| 247 | } |
| 248 | } |
| 249 | _ => { /* cannot squash incompatible types */ } |
| 250 | } |
| 251 | } |
| 252 | } |
| 253 | |
| 254 | impl Index<usize> for ClientBlockList { |
| 255 | type Output = BlockCell; |
| 256 | |
| 257 | fn index(&self, index: usize) -> &Self::Output { |
| 258 | &self.list[index] |
| 259 | } |
| 260 | } |
| 261 | |
| 262 | impl IndexMut<usize> for ClientBlockList { |
| 263 | fn index_mut(&mut self, index: usize) -> &mut Self::Output { |
| 264 | &mut self.list[index] |
| 265 | } |
| 266 | } |
| 267 | |
| 268 | pub(crate) struct ClientBlockListIter<'a>(std::slice::Iter<'a, BlockCell>); |
| 269 | |
| 270 | impl<'a> Iterator for ClientBlockListIter<'a> { |
| 271 | type Item = &'a BlockCell; |
| 272 | |
| 273 | fn next(&mut self) -> Option<Self::Item> { |
| 274 | self.0.next() |
| 275 | } |
| 276 | } |
| 277 | |
| 278 | pub(crate) struct ClientBlockListIterMut<'a>(std::slice::IterMut<'a, BlockCell>); |
| 279 | |
| 280 | impl<'a> Iterator for ClientBlockListIterMut<'a> { |
| 281 | type Item = &'a mut BlockCell; |
| 282 | |
| 283 | fn next(&mut self) -> Option<Self::Item> { |
| 284 | self.0.next() |
| 285 | } |
| 286 | } |
| 287 | |
| 288 | /// Block store is a collection of all blocks known to a document owning instance of this type. |
| 289 | /// Blocks are organized per client ID and contain a resizable list of all blocks inserted by that |
| 290 | /// client. |
| 291 | #[derive(PartialEq, Default)] |
| 292 | pub(crate) struct BlockStore { |
| 293 | clients: HashMap<ClientID, ClientBlockList, BuildHasherDefault<ClientHasher>>, |
| 294 | } |
| 295 | |
| 296 | pub(crate) type Iter<'a> = std::collections::hash_map::Iter<'a, ClientID, ClientBlockList>; |
| 297 | pub(crate) type IterMut<'a> = std::collections::hash_map::IterMut<'a, ClientID, ClientBlockList>; |
| 298 | |
| 299 | impl BlockStore { |
| 300 | /// Checks if block store is empty. Empty block store doesn't contain any blocks, neither active |
| 301 | /// nor tombstoned. |
| 302 | pub fn is_empty(&self) -> bool { |
| 303 | self.clients.is_empty() |
| 304 | } |
| 305 | |
| 306 | pub fn contains(&self, id: &ID) -> bool { |
| 307 | if let Some(clients) = self.clients.get(&id.client) { |
| 308 | id.clock < clients.clock() |
| 309 | } else { |
| 310 | false |
| 311 | } |
| 312 | } |
| 313 | |
| 314 | pub fn push_block(&mut self, block: Box<Item>) { |
| 315 | let id = block.id(); |
| 316 | match self.clients.entry(id.client) { |
| 317 | Entry::Occupied(mut e) => { |
| 318 | let list = e.get_mut(); |
| 319 | list.push(block.into()); |
| 320 | } |
| 321 | Entry::Vacant(e) => { |
| 322 | let list = e.insert(ClientBlockList::default()); |
| 323 | list.push(block.into()); |
| 324 | } |
| 325 | } |
| 326 | } |
| 327 | |
| 328 | pub fn push_gc(&mut self, gc: BlockRange) { |
| 329 | let id = gc.id; |
| 330 | let gc: BlockCell = GC::from(gc).into(); |
| 331 | match self.clients.entry(id.client) { |
| 332 | Entry::Occupied(mut e) => { |
| 333 | let list = e.get_mut(); |
| 334 | list.push(gc); |
| 335 | } |
| 336 | Entry::Vacant(e) => { |
| 337 | let list = e.insert(ClientBlockList::default()); |
| 338 | list.push(gc); |
| 339 | } |
| 340 | } |
| 341 | } |
| 342 | |
| 343 | /// Returns an iterator over the client and block lists pairs known to a current block store. |
| 344 | pub fn iter(&self) -> Iter<'_> { |
| 345 | self.clients.iter() |
| 346 | } |
| 347 | |
| 348 | /// Returns an iterator over the client and mutable block lists pairs known to a current block store. |
| 349 | pub fn iter_mut(&mut self) -> IterMut<'_> { |
| 350 | self.clients.iter_mut() |
| 351 | } |
| 352 | |
| 353 | /// Returns a state vector, which is a compact representation of the state of blocks integrated |
| 354 | /// into a current block store. This state vector can later be encoded and send to a remote |
| 355 | /// peers in order to calculate differences between two stored and produce a compact update, |
| 356 | /// that can be applied in order to fill missing update information. |
| 357 | pub fn get_state_vector(&self) -> StateVector { |
| 358 | let map = self |
| 359 | .clients |
| 360 | .iter() |
| 361 | .map(|(client_id, list)| (*client_id, list.clock())) |
| 362 | .collect(); |
| 363 | StateVector::new(map) |
| 364 | } |
| 365 | |
| 366 | pub(crate) fn get_client(&self, client_id: &ClientID) -> Option<&ClientBlockList> { |
| 367 | self.clients.get(client_id) |
| 368 | } |
| 369 | |
| 370 | pub(crate) fn get_client_mut(&mut self, client_id: &ClientID) -> Option<&mut ClientBlockList> { |
| 371 | self.clients.get_mut(client_id) |
| 372 | } |
| 373 | |
| 374 | /// Returns immutable reference to a block, given its pointer. Returns `None` if not such |
| 375 | /// block could be found. |
| 376 | pub(crate) fn get_block(&self, id: &ID) -> Option<&BlockCell> { |
| 377 | let clients = self.clients.get(&id.client)?; |
| 378 | clients.get_block(id.clock) |
| 379 | } |
| 380 | |
| 381 | pub(crate) fn get_block_mut(&mut self, id: &ID) -> Option<&mut BlockCell> { |
| 382 | let clients = self.clients.get_mut(&id.client)?; |
| 383 | clients.get_block_mut(id.clock) |
| 384 | } |
| 385 | |
| 386 | pub(crate) fn get_item(&self, id: &ID) -> Option<ItemPtr> { |
| 387 | let cell = self.get_block(id)?; |
| 388 | if let BlockCell::Block(item) = cell { |
| 389 | Some(ItemPtr::from(item)) |
| 390 | } else { |
| 391 | None |
| 392 | } |
| 393 | } |
| 394 | |
| 395 | /// Returns a block slice that represents a range of data within a particular block containing |
| 396 | /// provided [ID], starting from that [ID] until the end of the block. |
| 397 | /// |
| 398 | /// Example: *for a block `A:1..=5` and id `A:3`, the returned slice will represent `A:3..=5`*. |
| 399 | pub(crate) fn get_item_clean_start(&self, id: &ID) -> Option<ItemSlice> { |
| 400 | let ptr = self.get_item(id)?; |
| 401 | let offset = id.clock - ptr.id().clock; |
| 402 | Some(ItemSlice::new(ptr, offset, ptr.len() - 1)) |
| 403 | } |
| 404 | |
| 405 | /// Returns a block slice that represents a range of data within a particular block containing |
| 406 | /// provided [ID], starting from the beginning of the block until the that [ID] (inclusive). |
| 407 | /// |
| 408 | /// Example: *for a block `A:1..=5` and id `A:3`, the returned slice will represent `A:1..=3`*. |
| 409 | pub(crate) fn get_item_clean_end(&self, id: &ID) -> Option<ItemSlice> { |
| 410 | let ptr = self.get_item(id)?; |
| 411 | let block_id = ptr.id(); |
| 412 | let offset = id.clock - block_id.clock; |
| 413 | Some(ItemSlice::new(ptr, 0, offset)) |
| 414 | } |
| 415 | |
| 416 | /// Returns the last observed clock sequence number for a given `client`. This is exclusive |
| 417 | /// value meaning it describes a clock value of the beginning of the next block that's about |
| 418 | /// to be inserted. You cannot use that clock value to find any existing block content. |
| 419 | pub fn get_clock(&self, client: &ClientID) -> u32 { |
| 420 | if let Some(list) = self.clients.get(client) { |
| 421 | list.clock() |
| 422 | } else { |
| 423 | 0 |
| 424 | } |
| 425 | } |
| 426 | |
| 427 | /// Returns a mutable reference to block list for the given `client`. In case when no such list |
| 428 | /// existed, a new one will be created and returned. |
| 429 | pub(crate) fn get_client_blocks_mut(&mut self, client: ClientID) -> &mut ClientBlockList { |
| 430 | self.clients |
| 431 | .entry(client) |
| 432 | .or_insert_with(ClientBlockList::default) |
| 433 | } |
| 434 | |
| 435 | /// Given block pointer, tries to split it, returning a true, if block was split in result of |
| 436 | /// calling this action, and false otherwise. |
| 437 | pub fn split_block( |
| 438 | &mut self, |
| 439 | mut block: ItemPtr, |
| 440 | offset: u32, |
| 441 | encoding: OffsetKind, |
| 442 | ) -> Option<ItemPtr> { |
| 443 | let id = block.id().clone(); |
| 444 | let blocks = self.clients.get_mut(&id.client)?; |
| 445 | let index = blocks.find_pivot(id.clock)?; |
| 446 | let mut right = block.splice(offset, encoding)?; |
| 447 | let right_ptr = ItemPtr::from(&mut right); |
| 448 | blocks.insert(index + 1, right.into()); |
| 449 | |
| 450 | Some(right_ptr) |
| 451 | } |
| 452 | |
| 453 | pub(crate) fn split_block_inner(&mut self, block: ItemPtr, offset: u32) -> Option<ItemPtr> { |
| 454 | self.split_block(block, offset, OffsetKind::Utf16) |
| 455 | } |
| 456 | } |
| 457 | |
| 458 | impl std::fmt::Debug for ClientBlockList { |
| 459 | fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result { |
| 460 | std::fmt::Display::fmt(self, f) |
| 461 | } |
| 462 | } |
| 463 | |
| 464 | impl std::fmt::Display for ClientBlockList { |
| 465 | fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result { |
| 466 | f.debug_list().entries(self.list.iter()).finish() |
| 467 | } |
| 468 | } |
| 469 | |
| 470 | impl std::fmt::Debug for BlockStore { |
| 471 | fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result { |
| 472 | std::fmt::Display::fmt(self, f) |
| 473 | } |
| 474 | } |
| 475 | |
| 476 | impl std::fmt::Display for BlockStore { |
| 477 | fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result { |
| 478 | let mut s = f.debug_struct(""); |
| 479 | for (k, v) in self.clients.iter() { |
| 480 | s.field(&k.to_string(), v); |
| 481 | } |
| 482 | s.finish() |
| 483 | } |
| 484 | } |
| 485 | |
| 486 | pub(crate) struct Blocks<'a> { |
| 487 | current_client: std::vec::IntoIter<(&'a ClientID, &'a ClientBlockList)>, |
| 488 | current_block: Option<ClientBlockListIter<'a>>, |
| 489 | } |
| 490 | |
| 491 | impl<'a> Blocks<'a> { |
| 492 | fn new(update: &'a BlockStore) -> Self { |
| 493 | let mut client_blocks: Vec<(&'a ClientID, &'a ClientBlockList)> = |
| 494 | update.clients.iter().collect(); |
| 495 | // sorting to return higher client ids first |
| 496 | client_blocks.sort_by(|a, b| b.0.cmp(a.0)); |
| 497 | let mut current_client = client_blocks.into_iter(); |
| 498 | |
| 499 | let current_block = current_client.next().map(|(_, v)| v.iter()); |
| 500 | Blocks { |
| 501 | current_client, |
| 502 | current_block, |
| 503 | } |
| 504 | } |
| 505 | } |
| 506 | |
| 507 | impl<'a> Iterator for Blocks<'a> { |
| 508 | type Item = &'a BlockCell; |
| 509 | |
| 510 | fn next(&mut self) -> Option<Self::Item> { |
| 511 | if let Some(blocks) = self.current_block.as_mut() { |
| 512 | let block = blocks.next(); |
| 513 | if block.is_some() { |
| 514 | return block; |
| 515 | } |
| 516 | } |
| 517 | |
| 518 | if let Some(entry) = self.current_client.next() { |
| 519 | self.current_block = Some(entry.1.iter()); |
| 520 | self.next() |
| 521 | } else { |
| 522 | None |
| 523 | } |
| 524 | } |
| 525 | } |
| 526 | |
amadan